Interactive Plots
=================

With special plotting libraries like ``holoviews`` and ``datashader`` for big
data visualization as well as ``bokeh`` for interactiveness, we can use the
functionality of pyOpenMS to quickly create fully interactive views of
mass spectrometry data. Here we plot a full map of :term:`MS1` that can be
interactively zoomed-in if you execute the code in a notebook
(e.g. on Binder, see the button on top of the page).

.. code-block:: python
    :linenos:

    import pyopenms as oms
    import pandas as pd
    import numpy as np
    import datashader as ds
    import holoviews as hv
    import holoviews.operation.datashader as hd
    from holoviews.plotting.util import process_cmap
    from holoviews import opts, dim
    import sys

    hv.extension("bokeh")

    exp = oms.MSExperiment()  # type: PeakMap
    loader = oms.MzMLFile()
    loadopts = loader.getOptions()  # type: PeakFileOptions
    loadopts.setMSLevels([1])
    loadopts.setSkipXMLChecks(True)
    loadopts.setIntensity32Bit(True)
    loadopts.setIntensityRange(oms.DRange1(oms.DPosition1(5000), oms.DPosition1(sys.maxsize)))
    loader.setOptions(loadopts)
    loader.load("../../../src/data/BSA1.mzML", exp)
    exp.updateRanges()
    expandcols = ["RT", "mz", "inty"]
    spectraarrs2d = exp.get2DPeakDataLong(
        exp.getMinRT(), exp.getMaxRT(), exp.getMinMZ(), exp.getMaxMZ()
    )
    spectradf = pd.DataFrame(dict(zip(expandcols, spectraarrs2d)))
    spectradf = spectradf.set_index(["RT", "mz"])

    maxrt = spectradf.index.get_level_values(0).max()
    minrt = spectradf.index.get_level_values(0).min()
    maxmz = spectradf.index.get_level_values(1).max()
    minmz = spectradf.index.get_level_values(1).min()


    def new_bounds_hook(plot, elem):
        x_range = plot.state.x_range
        y_range = plot.state.y_range
        x_range.bounds = minrt, maxrt
        y_range.bounds = minmz, maxmz


    points = hv.Points(
        spectradf, kdims=["RT", "mz"], vdims=["inty"], label="MS1 survey scans"
    ).opts(
        fontsize={"title": 16, "labels": 14, "xticks": 6, "yticks": 12},
        color=np.log(dim("int")),
        colorbar=True,
        cmap="Magma",
        width=1000,
        height=1000,
        tools=["hover"],
    )

    raster = (
        hd.rasterize(
            points,
            cmap=process_cmap("blues", provider="bokeh"),
            aggregator=ds.sum("inty"),
            cnorm="log",
            alpha=10,
            min_alpha=0,
        )
        .opts(active_tools=["box_zoom"], tools=["hover"], hooks=[new_bounds_hook])
        .opts(  # weird.. I have no idea why one has to do this. But with one opts you will get an error
            plot=dict(
                width=800,
                height=800,
                xlabel="Retention time (s)",
                ylabel="mass/charge (Da)",
            )
        )
    )

    hd.dynspread(raster, threshold=0.7, how="add", shape="square")


Result:

.. image:: img/bokehms1.png


With this you can also easily create whole dashboards like the one
hosted `here <https://mybinder.org/v2/gh/OpenMS/pyopenms-docs/master+ipynb?urlpath=msbokehapps>`_ on a Binder instance.
If you are reading/executing this on Binder already, execute the next cell to get a link to your current instance.

.. code-block:: python
    :linenos:

    import os
    from IPython.display import Markdown as md

    md(
        "When you are in binder already, you can quickly open the app [here]({}/msbokehapps).".format(
            os.getenv("JUPYTERHUB_SERVICE_PREFIX")
        )
    )